AVTrack: New Audio-Visual Speaker Tracking Method

Arxiv CS.CV (computer vision)

AVTrack is a novel method for tracking active speakers using both audio and visual cues in complex human-centric scenes. The approach aims to enable fine-grained scene understanding for real-world applications.

Arxiv CS.CV (computer vision)·Yaoting WangYun ZhouZipei ZhangHenghui Ding

8 days ago